7 November 2026
GitHub censored me. I demand to be heard.
The plain facts: on 7 November 2026 GitHub's automated systems flagged my organization, marked me, its owner, as spammy, cancelled my builds, hid all 109 of my repositories, 91 of them public, and took this website down with them. The cause was my own AI coding agents, about twenty at once, writing to GitHub faster than its abuse filters allow. My machine account was suspended. An appeal is pending. So now the core of it lives on Radicle and on my own hardware, and GitHub holds only a signpost.
Nothing was deleted. Nothing I built went away. It just stopped being reachable, because all of it lived in one place I do not own.
What happened
All times are UTC, 7 November 2026.
- About 08:14 to 08:20. GitHub's automated systems flagged the aien-dev organization. Its page read: "The aien-dev organization has been flagged. Because of that, your organization is hidden from the public." It listed 109 repositories.
- From about 08:22. Public visitors got 404 Not Found for the repositories, among them CUDApp, omega-mv, omega-harness, omega-bridge and aien-sovereign-core-mv, and for www.aienos.com, which GitHub Pages served.
- Since then. The core public repositories are published on Radicle, the peer-to-peer git network, from my own node. This site now runs on my own hardware, a Raspberry Pi at home, instead of GitHub Pages.
Why
I caused it, and I will say so plainly. About twenty of my AI coding agents were working at once. Together they opened pull requests, pushed and re-ran CI faster than GitHub's abuse filters allow. The filters did what filters do. My machine account, aien-spark, was suspended, and the organization was hidden. I have appealed to GitHub Support, and the appeal is pending.
This is not a story about bad people. It is a story about a single point of failure.
What it proves
One company's automated switch hid 109 repositories, 91 of them public, and the website that explains them. The code was still on my disks. The people who wanted to read it could not reach it.
That is exactly the dependency sovereign computing exists to end. AienOS is built on one rule: intelligence should belong to the person running it, from the language to the machine. Code, history and a home on the web are part of that. If they live only on someone else's servers, they are yours for as long as those servers say so. I learned how fast that can change. So the rule now applies to me too.
Where it lives now
The one-command way in:
curl -fsSL https://www.aienos.com/si | shGitHub now holds only a signpost: the front door on GitHub (Omega source only).
The repositories below are published on Radicle from my own node, z6MkfZDue1TbaNFi3ZzcD9uiTeduPLC52ucdFVuLKZwexxhK. No company sits between you and the code. Install Radicle in two steps:
1. Download, verify and unpack the signed release for your system: https://radicle.xyz
2. rad authThen start your node and clone any repository below, for example this website:
rad node start
rad clone rad:z3iwnx2pktbGhnfBpJSsaFcTmKgHU28 public repositories are published from my own node so far. Each one clones with one command:
| Repository | Clone command |
|---|---|
aien-sovereign-core-mv | rad clone rad:z44GqrPVAPjZkhoD2MWVjqy18ighN |
omega-mv | rad clone rad:z6do9WDBhuSnvopDkgFdFr1yWA62 |
omega-harness | rad clone rad:z39gvwWsnkRZsAConQLjaKiNVLHfW |
aien-architecture-mv | rad clone rad:z37zcFYooiuCCZsf5L74eeFJYiwWM |
omega-bridge | rad clone rad:z3Hr9ZaiKHwwuNzPK5M8c7uw6pj7y |
aien-train | rad clone rad:zKrfazKqYNFuXDDkYdptxJicGLEq |
aegis-runtime-mv | rad clone rad:z6hhF5wGXCqvkcvxYdcBb12qm3FB |
aien | rad clone rad:zDTnULpiVMniD2etkbUozUP5mn6U |
aienix | rad clone rad:zV8VdbBDCPEZePcnBiseNyym3Wnk |
aienos.com | rad clone rad:z3iwnx2pktbGhnfBpJSsaFcTmKgHU |
aien-protocols-mv | rad clone rad:zzxgxYRc5tw6VwYXkQNLC2hqKkfb |
aien-receipts | rad clone rad:z4HXNamwwXyfN11ZBJXAQpdMKvojz |
aien-rig | rad clone rad:z2MPnsBZpJYeJJq39EXz1SYzLL7mc |
aienux | rad clone rad:z2Q6ob2gL6Qo5oUMpyehukayVF7mN |
benchmarks-mv | rad clone rad:z2nYCN9XZAPjNohqT8bz46rcs9zvZ |
interplane.io | rad clone rad:z399yKvnXcBQNkKoq86iBDgTvQ5QV |
interplane-mv | rad clone rad:zNhNbqYvgSBEHBq9vvE8xpbcBzGa |
physics-mv | rad clone rad:z49She8eifZpLd2aiSvk2pph3DMUo |
sovereign-intelligence | rad clone rad:z2JJz4RDfZZxKGBLn6UhmW5p9VJQj |
spark-rsi-mv | rad clone rad:z4AauBRPXzPNMiAoHZpRpyCee61A2 |
synapse | rad clone rad:zMGRREr2tmtRqpo4mMjpzkZdUGXN |
CUDApp | rad clone rad:z4RxD8vXigmWg61nM8miQ8DVPAXEP |
.github | rad clone rad:z36odegYfUoEoX8xwjJ4so68jb1W3 |
gpu-head-to-head | rad clone rad:z3ZSvMF8VFFGa5FyyKXGh4Vv6NBpS |
omega-blas | rad clone rad:z47VdC9EFsF4dFyH1QJ1C2JNJK4JS |
omega-gpu-driver | rad clone rad:z2xjD5kxUTTBNCryJ6rg9QXpemzei |
cuda-crack | rad clone rad:z631r2JJiVT8it4tST7RK24hpowi |
omega-mcp | rad clone rad:z3e8UBQVfwEcESJrcTWEbyMAHWFBZ |
To keep a copy alive on your own node too, run rad seed with the same RID.
And this site runs on my own hardware.
What I learned controlling the GPU in my DGX Spark without the CUDA stack is published too, with the code that reproduces each finding: the DGX Spark field manual.
Replicate it with any AI agent: AGENT-PROMPT.md
Join
Every copy makes the project harder to switch off. Run a Radicle node, clone the code, and seed what you clone: rad seed with a repository ID keeps it on your machine and serves it to the next person. That is how a project stops depending on any single company.
Tell me you are seeding, or that something is broken: drake@aienos.com, X @drakestapleton, or the contact page.